Volatiles released during volcanic activity can constrain magma dynamics, the transfer of volatiles from the mantle to the atmosphere, and mantle source geochemistry. Melt inclusions trapped at high temperature during crystal growth, and preserved as glass are a powerful tool to assess the pre-eruption behaviour of the magmatic volatile phase. Here we report the variations of CO2, H2O, Cl and S concentrations in melt inclusions in olivine crystals from the ongoing 1998 eruption of Piton de la Fournaise volcano. This volcano is one of the most active hot spot volcanoes on Earth. Located in the Indian Ocean, on Réunion Island, it is related to the hot spot that produced the Deccan Trapps some 65 Ma ago. The 1998 eruption was extremely carefully followed and studied by the Volcanological Observatory of Réunion (Institut de Physique du Globe). Lavas were regularly sampled and quenched in water. The investigated melt inclusions were selected from these quenched samples, to preclude any chemical exchange between the trapped silicate melt and the host olivine. We focused our attention on the products simultaneously erupted along fractures located south and north of the Dolomieu crater. These fractures are called Piton Hudson and Piton Kapor.
Primary melt inclusions were polished on two faces and analysed for CO2 and H2O, using FTIR microspectroscopy (LPS-Saclay, BGI). The major element compositions were determined for the both olivine and their melt inclusions by electron microprobe (BGI and Camparis Paris VI). The Cl and S contents, and the relative proportions of sulphur dissolved as sulphate were also measured.
The Kapor melt inclusions are chemically evolved (MgO from 6 to 7.7 wt% in olivine Fo78-85) and degassed (low C and H2O contents). The Hudson melt inclusions in olivines Fo82-87 are MgO-rich (7.2 to 10.2 wt%) with unusually high volatile contents of up to 1.6 wt% H2O and up to 1200 ppm carbon. The olivines (Fo89) in Hudson samples contain high density secondary CO2-rich fluid inclusions. From the volatile contents of the melt inclusions, the total fluid pressure should vary from 6 kbars to few hundred of bars. It appears that the Hudson olivines come from more than 20 km depth, whereas the olivines erupted by Kapor come from shallow levels. These results suggest a vertical chemical zonation in the system. Apparently, olivine crystals were transported by the ascending magma from different levels of the volcano, while the eruptive fractures opened in the uppermost parts of the volcano. Aside from small lava fountains (10 meter high) during a few days, no major degassing processes occured at Piton Hudson, but high lava fountains (50 m) were reported for Piton Kapor during March 1998 and after. For this month, and using the maximum volatile contents recorded in the melt inclusions and the highest measured value of magma flux (30 m3/sec), the volatile emission has been calculated to be 0.4 ton/day CO2 and about 1.3 ton/day H2O. In addition, the maximal depth for seismic signals recorded before and during the 1998 eruption was less than 5 km.
It appears that the magma batches are losing H2O and CO2 slowly during their ascent from more than 20 km depth to the surface, without seismic activity. There is no evidence of degassing of S, and CO2 at the surface, prior to the eruption. Most of the volatile constituents are probably completely diluted in aquifers.